A soil quality monitoring system for uncultivated land

Abstract

This invention relates to the field of agricultural environmental monitoring and intelligent sensor technology, specifically disclosing a soil quality monitoring system for uncultivated wasteland. The system includes a multi-parameter soil sensing array, a microbial metabolic response activation module, a metabolic response signal capture and analysis module, a soil biological activity index calculation engine, and a soil quality comprehensive evaluation and source tracing decision unit. By applying standardized metabolic substrates in situ and capturing multi-dimensional response signals, the system calculates and quantifies the biological activity index, and then integrates physicochemical parameters for comprehensive evaluation and source tracing diagnosis. This enables dynamic, in-situ quantitative assessment and in-depth soil quality analysis of the functional activity of soil microorganisms in uncultivated wasteland.

Description

Technical Field

[0001] This invention belongs to the field of agricultural environmental monitoring and intelligent sensor technology, specifically relating to a soil quality monitoring system for uncultivated wasteland. Background Technology

[0002] Soil quality monitoring is a core technology in modern agriculture, ecological restoration, and land resource management. It aims to provide a scientific basis for sustainable land use by assessing the physical, chemical, and biological properties of soil. Soil biological activity, as a key indicator of soil health and ecological function, is directly related to nutrient cycling, pollutant degradation, and crop productivity.

[0003] The core objective of soil quality monitoring for uncultivated wasteland is to rapidly and accurately assess the initial fertility and ecological potential of this special type of land, providing fundamental data for subsequent land reclamation, vegetation restoration, or agricultural planning. Traditional monitoring methods rely on laboratory chemical analysis, which, while highly accurate, is time-consuming and labor-intensive, making it difficult to meet the needs of large-scale, dynamic monitoring.

[0004] Existing technologies have attempted to incorporate various soil sensors for rapid on-site detection of soil moisture, temperature, pH, and some nutrient content. However, these sensors primarily focus on soil physicochemical parameters, and their methods have significant limitations in reflecting bioactivity indicators that represent the core functions of the soil ecosystem, particularly the composition and functional state of microbial communities. Soil bioactivity is difficult to quantify directly using conventional sensors, and the metabolic functions, diversity, and environmental responses of microbial communities cannot be effectively captured and characterized. This results in superficial assessments of the ecological quality of uncultivated soils, failing to reveal their true biofertility and restoration potential, and thus lacking crucial evidence for developing scientific land management strategies. Summary of the Invention

[0005] This invention provides a soil quality monitoring system for uncultivated wasteland, in order to resolve the technical contradiction in existing soil quality monitoring systems that are difficult to directly, dynamically, and in-situ quantify and assess the biological activity and microbial functional status of uncultivated wasteland soil.

[0006] The technical solution of the present invention is a soil quality monitoring system for uncultivated wasteland, the system comprising: A multi-parameter soil sensing array is deployed at monitoring points in target uncultivated wasteland to perform in-situ synchronous acquisition of basic soil physicochemical parameters. The microbial metabolic response activation module is used to apply a set of preset standardized metabolic substrate stimuli to the same soil microdomain located by the soil multi-parameter sensing array. The metabolic response signal capture and analysis module is used to monitor and record in real time the multidimensional metabolic response kinetic curves generated in the soil microdomain under the stimulation of standardized metabolic substrates. The soil bioactivity index calculation engine is used to generate a set of quantitative bioactivity indices that comprehensively characterize the functional activity and metabolic potential of soil microbial communities based on the multi-dimensional metabolic response dynamic curves obtained by the metabolic response signal capture and analysis module through feature extraction and model fusion calculation. The soil quality comprehensive evaluation and source tracing decision unit is used to integrate the basic physicochemical parameters collected by the soil multi-parameter sensing array with the quantitative bioactivity index generated by the soil bioactivity index calculation engine, perform multi-level data fusion and model inference, and output a comprehensive quality report including soil health level, limiting factor diagnosis and ecological restoration potential assessment.

[0007] Furthermore, the soil multi-parameter sensing array consists of multiple heterogeneous sensing probes integrated in a star-shaped topology into the end sensing head of a rigid probe. Each sensing probe is evenly distributed in space in a circular pattern, and its sensing end face is on the same horizontal plane to ensure simultaneous measurement of the same soil micro-domain.

[0008] The soil multi-parameter sensing array specifically integrates a soil volumetric water content sensor, a soil temperature sensor, a soil pH sensor, a soil electrical conductivity sensor, and a soil redox potential sensor.

[0009] All sensor probes are connected to the data acquisition and preprocessing circuit board inside the probe rod via internal shielded cables. The preprocessing circuit board is responsible for synchronously sampling, analog-to-digital conversion, temperature compensation, and preliminary filtering of the analog signals output by each sensor, generating a data packet of physicochemical basic parameters with timestamp synchronization, and uploading it through the first wireless communication module integrated inside the probe rod.

[0010] Furthermore, the microbial metabolic response excitation module is integrated at the center of the sensing head of the rigid probe, and includes a miniature multi-chamber liquid storage tank, a set of miniature precision injection pumps, and a ring-shaped microdialysis array.

[0011] The miniature multi-chamber reservoir is independently encapsulated with at least four different types of standardized metabolic substrate concentrates, with each substrate concentrate corresponding to an independent reservoir.

[0012] The number of miniature precision injection pumps is the same as the number of reservoirs. The inlet of each injection pump is connected to the corresponding reservoir via a medical-grade silicone tube, and the outlet is connected to the corresponding independent microchannel on the annular microdialysis array.

[0013] The annular microdialysis array is made of biocompatible material with nanoscale pores uniformly distributed on its surface. The annular microdialysis array is arranged coaxially around the central area of ​​the heterogeneous sensing probe and is in close contact with the soil.

[0014] The working process of the microbial metabolic response activation module is triggered by the central controller. Specifically, the central controller drives the designated micro precision injection pumps sequentially or in parallel according to the preset activation protocol to inject a standardized metabolic substrate concentrate of a specific type and volume into the corresponding annular microdialysis array microchannel at a constant flow rate. The substrate molecules diffuse slowly and uniformly into the surrounding soil microdomains through the nanopores.

[0015] Furthermore, the metabolic response signal capture and analysis module includes a gaseous metabolite monitoring submodule, a solution ion monitoring submodule, and a thermodynamic monitoring submodule.

[0016] The gas metabolite monitoring submodule consists of a high-sensitivity carbon dioxide sensor and a methane sensor, both of which are connected to a miniature gas collection hood via a gas pipeline. The gas collection hood is retractably installed above the probe sensing head and descends and seals the soil surface containing the micro-percolation array during the excitation phase, forming a temporary micro-gas chamber for real-time monitoring of the carbon dioxide and methane concentration changes produced by soil microbial respiration.

[0017] The solution ion monitoring submodule consists of a set of ion-selective field-effect transistor sensors. The set of ion sensors is embedded in a specific section of the annular microdialysis array to monitor in real time the dynamic change curve of the concentration of specific ions in the soil solution under substrate stimulation.

[0018] The thermodynamic monitoring submodule consists of a high-precision miniature thermopile sensor, whose thermosensitive nodes are in direct contact with the soil micro-domain, and is used to monitor the temperature change curve of the soil micro-domain caused by microbial metabolic activities in real time.

[0019] The metabolic response signal capture and analysis module is activated at the same time as the microbial metabolic response excitation module starts. It synchronously records gas concentration, ion concentration and temperature change data at a sampling frequency of no less than once per second, and integrates them into a time-aligned multi-dimensional metabolic response kinetic curve data packet, which is then uploaded through the second wireless communication module.

[0020] Furthermore, the soil bioactivity index calculation engine receives the multidimensional metabolic response kinetic curve, and its calculation process follows the following steps: First, curve preprocessing is performed, and each dynamic curve is filtered by moving average to remove high-frequency noise, and the time axis is aligned based on the excitation start time.

[0021] Next, feature extraction is performed to extract a set of preset time-domain and frequency-domain feature parameters from each preprocessed dynamic curve. The time-domain feature parameters include the maximum response intensity of the curve, the time required to reach 50% of the maximum intensity, the total area under the curve, and the time required for the response to decay to 10% of the peak value. The frequency-domain feature parameters include the amplitude of the main frequency component obtained by fast Fourier transform.

[0022] Then, feature fusion and index calculation are performed. Feature parameters from different monitoring sub-modules are input into a pre-trained multilayer perceptron neural network model. The number of input layer nodes of the perceptron neural network model is equal to the total number of feature parameters. The hidden layer contains two fully connected layers, and the number of output layer nodes is three, corresponding to three core bioactivity indices: basal metabolic activity index, substrate utilization diversity index, and metabolic function stability index.

[0023] The basal metabolic activity index is determined by the weighted sum of the total area under the carbon dioxide production curve and the maximum response intensity, and mainly reflects the overall respiratory metabolic intensity of the microbial community.

[0024] The substrate utilization diversity index is obtained by calculating the cosine similarity matrix between the feature vectors of response curves to different types of standardized metabolic substrates, and further calculating the information entropy of the cosine similarity matrix. The higher the entropy value, the more diverse the types of substrates that the microbial community can utilize.

[0025] The metabolic function stability index is calculated by analyzing the rate of change of the decay time constant and recovery rate of the microbial response curve when different substrates are applied continuously or alternately. The lower the rate of change, the more stable the metabolic function of the microbial community.

[0026] Furthermore, the data fusion and model reasoning process performed by the comprehensive soil quality evaluation and source tracing decision-making unit is as follows: The soil quality comprehensive evaluation and source tracing decision-making unit has a built-in soil quality knowledge graph, which stores complex association rules between soil physicochemical parameters, biological activity index and soil health status, ecological function and limiting factors in the form of triples.

[0027] First, data fusion is performed, where the received physicochemical basic parameter data package and the quantitative biological activity index data package are spatiotemporally matched and normalized to form a unified soil feature vector for the current monitoring point.

[0028] Next, feature layer fusion and inference are performed. The soil feature vector is input into the inference model based on the coupling of knowledge graph and Bayesian network. The operation process of the inference model is as follows: using the soil feature vector as observation evidence, activating the relevant association rules in the knowledge graph, calculating the posterior probability of different soil health level hypotheses, and determining the level with the highest posterior probability as the soil health level of this monitoring point.

[0029] Meanwhile, the data fusion and model executed by the soil quality comprehensive evaluation and source tracing decision-making unit, based on the thresholds and logical relationships defined in the knowledge graph, diagnoses the main limiting factors that cause the soil health level to not reach the optimal level. The limiting factors are classified as physical factors, chemical factors, or biological factors.

[0030] Finally, a decision-making integration and potential assessment were conducted. By combining the soil health level, the results of the limiting factor diagnosis, and the time series trend of historical monitoring data, a linear weighted decision model was used to calculate the comprehensive score of the ecological restoration potential of this uncultivated wasteland, and a structured comprehensive quality report was generated. The report clearly includes the soil health level, a specific list of limiting factors, the ecological restoration potential score, and targeted preliminary management recommendations.

[0031] Furthermore, the standardized metabolic substrate concentrate includes at least a glucose solution, a protein hydrolysate solution, a cellulose derivative solution, and a phenolic compound solution, which are used to stimulate the metabolism of readily degradable carbon sources, nitrogen sources, difficult-to-degrade carbon sources, and aromatic compounds by soil microorganisms, respectively.

[0032] Furthermore, the system adopts a hybrid power supply mode that combines a solar power supply unit with a high-capacity lithium-ion battery pack. The solar power supply unit includes a foldable photovoltaic panel and a maximum power point tracking charge controller, which is used to power the system during the day and charge the lithium-ion battery pack. The lithium-ion battery pack is used to provide continuous power at night or when there is insufficient sunlight, ensuring that the system can perform long-term unattended continuous or periodic monitoring.

[0033] Furthermore, the system is deployed as a networked monitoring system, with individual monitoring points constituting monitoring nodes, and multiple monitoring nodes forming a wireless sensor network through a self-organizing network protocol. Each monitoring node aggregates the processed data to a regional gateway node, which then uploads the data to a cloud server for storage, in-depth analysis, and visualization via long-distance wireless communication technology.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention constructs a complete technical chain from in-situ excitation and multi-dimensional signal capture to intelligent analysis and comprehensive evaluation, achieving for the first time a direct, dynamic, and in-situ quantitative assessment of the functional activity of soil microbial communities in uncultivated wasteland. By integrating a microbial metabolic response excitation module and a multi-dimensional metabolic response signal capture and analysis module, the system transforms invisible microbial metabolic processes into real-time monitorable physical signal dynamic curves, overcoming the limitations of traditional sensors that can only detect static physicochemical parameters. The soil bioactivity index calculation engine employs a method that integrates feature extraction and neural network models to extract core indices characterizing basal metabolic intensity, substrate utilization diversity, and functional stability from complex dynamic curves, enabling the assessment of soil biofertility and ecological function to move from qualitative description to precise quantification.

[0035] 2. The comprehensive soil quality assessment and source tracing decision-making unit integrates multi-source data on physicochemical and biological activities, and uses knowledge graphs and Bayesian networks for reasoning. It not only outputs a comprehensive soil health level but also accurately diagnoses limiting factor types and provides an assessment of ecological restoration potential. This provides unprecedented in-depth decision support for the scientific reclamation, precise improvement, and ecological management of uncultivated wasteland. The entire system design achieves automated, networked, and intelligent operation, significantly improving the efficiency and depth of soil quality monitoring in large-scale uncultivated wasteland, and possesses significant practical value and broad prospects for widespread application. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the overall technical architecture of the soil quality monitoring system for uncultivated wasteland proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of microbial metabolic response excitation and multidimensional signal capture and analysis in this invention; Figure 3 This is a logical flow diagram of the soil biological activity index calculation engine in this invention; Figure 4 This is a data fusion and reasoning framework diagram of the soil quality comprehensive evaluation and source tracing decision unit in this invention; Figure 5 This is a schematic diagram of the hierarchical interaction and data flow of the networked monitoring system in this invention. Detailed Implementation

[0037] This invention provides a soil quality monitoring system for uncultivated wasteland, the overall technical architecture of which is shown in the attached figure. Figure 1As shown in the figure. This system integrates five functional modules: in-situ sensing, active stimulation, multi-dimensional response capture, intelligent index calculation, and comprehensive decision evaluation. It constructs a closed-loop, dynamic, and quantifiable soil bioactivity assessment system. Deployed at representative monitoring points in target uncultivated land, the system can complete the entire process—from collecting soil physicochemical parameters, stimulating microbial metabolic functions, capturing metabolic response signals in real time, to generating bioactivity indices and comprehensively diagnosing soil quality—under unmanned intervention, achieving a deep characterization of soil ecological health. The following will be combined with the attached... Figures 1 to 5 The specific implementation methods of each component of the system are described in detail.

[0038] The core hardware of the system is a rigid probe, the end of which forms a highly integrated sensing head. The sensing head integrates all sub-components of a soil multi-parameter sensing array, a microbial metabolic response excitation module, and a metabolic response signal capture and analysis module. All modules are spatially tightly coupled, ensuring that all collected data originates from the same soil microdomain, thereby guaranteeing the spatiotemporal consistency of subsequent data fusion and model inference.

[0039] The soil multi-parameter sensing array is arranged in a star topology on the outer edge of the sensing head. Specifically, the array consists of five heterogeneous sensing probes: a soil volumetric water content sensor, a soil temperature sensor, a soil pH sensor, a soil conductivity sensor, and a soil redox potential sensor. These five probes are evenly distributed circumferentially, with their sensing faces strictly aligned in the same horizontal plane. This ensures that after the probe is inserted into the soil, all sensors simultaneously contact and measure the exact same soil microenvironment. Each probe is connected to a data acquisition and preprocessing circuit board embedded inside the probe via an internal shielded cable.

[0040] The preprocessing circuit board integrates a high-precision analog-to-digital converter, a temperature compensation algorithm unit, and a digital filter. It can synchronously sample the analog signals output from each sensor (sampling frequency no less than 10 times per second) and perform cross-compensation correction based on the measured soil temperature after analog-to-digital conversion to eliminate measurement deviations caused by temperature drift. After preliminary filtering and timestamping, a set of time-synchronized physicochemical basic parameter data packets is formed, including volumetric water content (unit: m³). 3 / m 3 The data package contains the following parameters: temperature (°C), pH value (dimensionless), conductivity (μS / cm), and redox potential (mV). These physicochemical parameters are periodically uploaded to the central controller or regional gateway node via a first wireless communication module integrated within the probe, using Bluetooth Low Energy or ZigBee protocols.

[0041] Combined with appendix Figure 2The microbial metabolic response activation module is located at the geometric center of the sensing head, forming a concentric layout with the surrounding soil multi-parameter sensing array. The module consists of three parts: a micro-multi-chamber storage tank, a set of micro-precision injection pumps, and a ring-shaped microdialysis array. The micro-multi-chamber storage tank is encapsulated in medical-grade polytetrafluoroethylene (PTFE) material and internally divided into four completely isolated storage chambers, each independently storing a standardized metabolic substrate concentrate. According to the system design, these four substrates are: a 0.1 mol / L glucose solution (for activating readily biodegradable carbon source metabolism), a 5 g / L protein hydrolysate solution (for activating nitrogen source metabolism), a 2 g / L sodium carboxymethyl cellulose solution (as a cellulose derivative, for activating recalcitrant carbon source metabolism), and a 0.5 mmol / L phenol solution (for activating aromatic compound degradation metabolism). Each substrate has a volume of at least 2 ml, supporting at least three rounds of activation experiments within a single complete monitoring cycle.

[0042] The system comprises four miniature precision infusion pumps, each corresponding to a reservoir. The inlet of each pump is connected to its corresponding reservoir via a 0.5 mm inner diameter medical-grade silicone tube, while the outlet is connected to a dedicated microchannel inlet on the annular microdialysis array. The annular microdialysis array is made of polyethersulfone biocompatible membrane material, is ring-shaped, with an outer diameter of 25 mm, an inner diameter of 15 mm, and a thickness of 1 mm. Its surface is plasma-treated to form a uniform microporous structure with an average pore size of 50 nanometers.

[0043] The annular microdialysis array is coaxially arranged around the center of a circular area enclosed by five heterogeneous sensing probes, and directly and closely adheres to the surrounding soil particles after the probes are inserted into the soil. When the central controller issues an excitation command, according to a preset excitation protocol (e.g., sequentially applying four substrates, each with an injection volume of 50 μL, a constant flow rate of 2 μL / s, and an interval of 30 minutes), the corresponding micro-precision injection pumps are sequentially activated, pushing the designated substrate concentrate into the designated microchannels of the annular microdialysis array. Driven by the concentration gradient, the substrate molecules slowly and uniformly diffuse through the nanopores into the surrounding soil microdomain with a radius of approximately 2 cm, forming a local high-concentration stimulation zone, thereby effectively activating the microbial population with corresponding metabolic capabilities within the local high-concentration stimulation zone.

[0044] Simultaneously, the metabolic response signal capture and analysis module is activated, beginning to record multidimensional physicochemical signal changes triggered by microbial metabolic activity. This module consists of three sub-modules: a gaseous metabolite monitoring sub-module, a solution ion monitoring sub-module, and a thermodynamic monitoring sub-module.

[0045] The gas metabolite monitoring submodule includes a high-sensitivity infrared carbon dioxide sensor (detection limit of 1 ppm) and a catalytic combustion methane sensor (detection limit of 0.1 ppm). Both are connected to a miniature gas collection hood via a 2 mm inner diameter PTFE gas conduit. The gas collection hood employs a shape memory alloy drive mechanism, normally retracting into the side wall of the probe; during the excitation phase, a telescopic mechanism is triggered by the central controller, causing it to descend vertically approximately 3 cm, tightly pressing against the soil surface to form a temporary, sealed micro-gas chamber with a volume of approximately 50 cubic centimeters, completely covering the annular micro-osmosis array and the surrounding soil area. Within this chamber, carbon dioxide and methane gases produced by microbial respiration are captured and analyzed in real time, generating concentration-time curves. The sampling frequency is set to once per second, with continuous monitoring for 60 minutes until the gas concentration stabilizes.

[0046] The solution ion monitoring submodule consists of four ion-selective field-effect transistor sensors, each targeting ammonium ions (NH4+). + ), nitrate ions (NO3) - ), phosphate ions (PO4) 3- ) and potassium ions (K + These sensors are embedded in specific segments of a ring-shaped microdialysis array, with their sensitive membranes directly exposed to the soil solution. During substrate stimulation, microbial metabolic activities alter the ionic composition and concentration of the local soil solution; for example, protein hydrolysate degradation releases ammonium ions, and nitrification generates nitrate ions. The ion-selective field-effect transistor sensor detects changes in channel current to reflect the dynamic evolution of target ion activity in real time, generating a corresponding ion concentration-time response curve. The sampling frequency is also once per second.

[0047] The thermodynamic monitoring submodule is implemented using a high-precision miniature thermopile sensor. Its thermistor node has a diameter of 1 mm and is directly embedded in the soil 2 mm below the center of the annular micro-dialysis array. Microbial metabolic processes, accompanied by energy release, result in a weak but measurable temperature rise in local soil micro-domains (typically ranging from 0.01℃ to 0.1℃). The thermopile sensor records the temperature rise process twice per second, forming a temperature-time kinetic curve. The rise rate, peak value, and decay rate of the temperature-time kinetic curve are all closely related to the intensity of microbial metabolism.

[0048] The raw data collected by the above three sub-modules undergoes time alignment, unit unification, and preliminary calibration in the signal conditioning circuit inside the probe, integrating it into a multi-dimensional metabolic response kinetic curve data packet with precise timestamp synchronization. This multi-dimensional metabolic response kinetic curve data packet is then uploaded to the central controller via a second wireless communication module (physically isolated from the first wireless communication module to avoid interference), for subsequent processing by the soil biological activity index calculation engine.

[0049] Combined with appendix Figure 3 After receiving the multi-dimensional metabolic response kinetic curve data packet, the soil biological activity index calculation engine performs a three-stage processing flow: curve preprocessing, feature extraction, feature fusion and index calculation.

[0050] In the curve preprocessing stage, each original kinetic curve (including CO2 concentration, CH4 concentration, NH4 concentration, etc.) is processed. + Concentration, NO3 - Concentration, PO4 3- Concentration, K + Seven curves (concentration, temperature, etc.) were analyzed using a 5-second moving average filter to suppress high-frequency electronic noise and environmental disturbances. Subsequently, the time axis of all curves was realigned using the standardized time of substrate injection completion as the time zero point to ensure that subsequent feature extraction was based on a unified time reference.

[0051] In the feature extraction stage, four time-domain feature parameters and one frequency-domain feature parameter are extracted from each preprocessed curve, totaling 35 feature parameters. The time-domain features include: maximum response intensity (i.e., curve peak value), and the time required to reach 50% of the maximum intensity. ), total area under the curve (AUC, calculated using the trapezoidal integral method), and time required for the response to decay to 10% of its peak value. Frequency domain characteristics are represented by the amplitude of the dominant frequency component (i.e., the frequency component with the largest amplitude) after performing a Fast Fourier Transform on the curve. For example, for The CO2 response curve, if it rises rapidly after glucose injection and reaches a peak of 800 ppm within 15 minutes, then the maximum response intensity is 800. The time is 7.5 minutes, and AUC is the integral value (unit: ppm·min). It may last for 45 minutes; its main frequency component may appear around 0.0003Hz, and the amplitude reflects the periodicity of the response.

[0052] In the feature fusion and index calculation stage, 35 feature parameters are input into a pre-trained multilayer perceptron neural network model. The multilayer perceptron neural network model has the following structure: 35 nodes in the input layer, 64 fully connected nodes in the first hidden layer (with ReLU activation function), 32 fully connected nodes in the second hidden layer (with ReLU activation function), and 3 nodes in the output layer, corresponding to the basal metabolic activity index (BMAI), substrate utilization diversity index (SUDI), and metabolic function stability index (MFSI), respectively.

[0053] The basal metabolic activity index is mainly determined by the weighted average of the AUC and the maximum response intensity of the CO2 response curve. The weighting coefficients are determined through regression analysis of historical data, with a typical value of 70% for AUC and 30% for the peak value. The basal metabolic activity index ranges from 0 to 100, with higher values ​​indicating stronger overall respiratory and metabolic capacity of the microorganisms.

[0054] The substrate diversity index is obtained by calculating the cosine similarity matrix between response feature vectors under different substrate excitations, and further calculating the information entropy of the cosine similarity matrix. Let the response feature vectors for the four substrates be as follows: , , , (Each vector is 35-dimensional), then the elements of the similarity matrix S are... After normalizing the S matrix row by row, calculate the Shannon entropy for each row. The average value is then taken to obtain the overall entropy value. This entropy value is linearly mapped to the range of 0 to 100, which is the SUDI. The higher the entropy value, the greater the difference in the response patterns of the microbial community to different types of substrates, and the richer the functional diversity.

[0055] The metabolic function stability index is calculated by analyzing the rate of change of the decay time constant and recovery rate of the response curve in a series of excitation experiments. For example, in two excitations of the same substrate with a 30-minute interval, if the first... If the first time is 40 minutes and the second time is 42 minutes, then the rate of change of the decay time constant is (42-40) / 40=5%; if the peak value drops from 800 to 780, then the recovery rate is 780 / 800=97.5%. The weighted average of the rates of change of multiple such indicators, taking its reciprocal and mapping it to 0 to 100, yields the MFSI. The smaller the rate of change, the higher the MFSI, indicating stronger system immunity to disturbances and greater functional stability.

[0056] The above three indices together constitute a quantitative bioactivity index data package, which is uploaded to the soil quality comprehensive evaluation and traceability decision-making unit.

[0057] Combined with appendix Figure 4 The soil quality comprehensive evaluation and source tracing decision-making unit performs three-layer fusion reasoning: data layer fusion, feature layer fusion and reasoning, and decision layer fusion and potential assessment.

[0058] In the data fusion stage, the five physicochemical parameters from the soil multi-parameter sensing array are spatiotemporally matched with the three bioactivity indices from the computing engine. Since all data originate from the same probe and the same time window, only dimensional normalization is required. Normalization employs the min-max scaling method, mapping each parameter to the [0,1] interval to form an 8-dimensional soil feature vector for the current monitoring point. .

[0059] In the feature layer fusion and inference stage, the system incorporates a soil quality knowledge graph. This knowledge graph stores over 2000 association rules in triplet (subject, relation, object) format, such as (soil pH < 5.0, leading to aluminum toxicity inhibiting microbial activity), (BMAI > 70, associated with high organic matter mineralization rate), (electrical conductivity > 2000 μS / cm, belonging to salinization limiting factors), etc. The inference model employs a coupled architecture of knowledge graph and Bayesian network. First, the feature vector X is input into the Bayesian network as observational evidence; second, rule nodes related to each component of X in the knowledge graph are activated; then, based on a conditional probability table, the posterior probability P(H|X) of five preset soil health levels (Level I: Excellent, Level II: Good, Level III: Medium, Level IV: Poor, Level V: Inferior) is calculated. The level with the highest posterior probability is the judgment result. Simultaneously, the model traverses the limiting factor rule base in the knowledge graph to identify parameter items whose actual observed values ​​exceed the threshold. For example, if pH=4.2 (below the threshold of 5.0) and BMAI=30 (significantly low), then “strong acidity” is diagnosed as a chemical limiting factor; if SUDI=20 (far below the benchmark value of 60), then “single microbial function” is diagnosed as a biological limiting factor.

[0060] During the decision-making integration and potential assessment phase, the system retrieves historical time series data (if available) from the monitoring points to analyze the changing trends of each indicator. For example, if BMAI shows a monthly upward trend of 5% over the past 6 months, it is considered positive succession. Combining the current health level, the number and type of limiting factors, and the historical trend slope, a comprehensive score for ecological restoration potential is calculated using a linear weighted decision model. :

[0061] in, The quantitative score for health level (Level I = 100, Level II = 80, Level III = 60, Level IV = 40, Level V = 20). The severity index for limiting factors (values ​​from 0 to 1; the more factors and the more critical they are, the larger L becomes). Historical trend coefficient (positive for upward trend, negative for downward trend, range -0.2 to +0.2), weight. =0.5, =0.3, =0.2. The R value ranges from 0 to 100. ≥80 indicates high potential, 60≤ <80 represents medium potential. <60 indicates low potential.

[0062] Finally, the system generates a structured comprehensive quality report, including: soil health level (e.g., "Level III: Medium"), a list of limiting factors (e.g., "Chemical: Strongly acidic (pH=4.2); Biological: Single microbial function (SUDI=20)"), and an ecological restoration potential score (e.g., " =65, medium potential) and targeted preliminary management recommendations (such as "It is recommended to apply lime to adjust the pH to above 6.0 and inoculate with multifunctional microbial agents to enhance substrate utilization diversity").

[0063] The entire system employs a hybrid power supply mode to support long-term field operation. The solar power unit consists of a foldable monocrystalline silicon photovoltaic panel (5 watts peak power) and a maximum power point tracking (MPPT) charger. During the day, the photovoltaic panel directly powers the system while simultaneously charging the built-in high-capacity lithium-ion battery pack (12 volts nominal, 10 amp-hour capacity) via the charger. At night or during prolonged cloudy or rainy weather, the system automatically switches to battery power, capable of maintaining full-load operation for 72 consecutive hours or undergoing 30-day periodic monitoring (performing a complete monitoring cycle every 24 hours).

[0064] Combined with appendix Figure 5 The system can be deployed as a networked monitoring system. Each monitoring point constitutes an independent monitoring node, possessing complete sensing, activation, computing, and communication capabilities. Multiple monitoring nodes (typically 4 to 9 per square kilometer) form a wireless sensor network via self-organizing network protocols (such as LoRaWAN or Thread). Each node encrypts and uploads processed physicochemical parameters, biological activity indices, and a summary of the comprehensive quality report to the regional gateway node. The regional gateway node is equipped with a 4G / 5G or NB-IoT communication module, which uniformly uploads the aggregated data to a cloud server. The cloud platform is responsible for long-term data storage, cross-regional comparative analysis, trend prediction, and visualization, supporting agricultural management departments in conducting large-scale surveys of uncultivated wasteland resources and ecological restoration planning.

[0065] In summary, this embodiment, through highly integrated hardware design, standardized metabolic activation protocol, multi-dimensional response signal capture mechanism, and intelligent analysis model based on neural networks and knowledge graphs, has for the first time achieved in-situ, dynamic, and quantitative assessment of the functional status of soil microorganisms in uncultivated wasteland. This breakthrough overcomes the limitations of traditional soil monitoring that relies solely on static physicochemical indicators, providing solid technical support for the scientific evaluation and sustainable utilization of wasteland resources.

Claims

1. A soil quality monitoring system for uncultivated wasteland, characterized in that, include: A multi-parameter soil sensing array is deployed at monitoring points in target uncultivated wasteland to perform in-situ synchronous acquisition of basic soil physicochemical parameters. The microbial metabolic response activation module is used to apply a set of preset standardized metabolic substrate stimuli to the same soil microdomain located by the soil multi-parameter sensing array. The metabolic response signal capture and analysis module is used to monitor and record in real time the multidimensional metabolic response kinetic curves generated in the soil microdomain under the stimulation of the standardized metabolic substrate. The soil bioactivity index calculation engine is used to generate a set of quantitative bioactivity indices based on the multidimensional metabolic response dynamic curves obtained by the metabolic response signal capture and analysis module, through feature extraction and model fusion calculation. The soil quality comprehensive evaluation and source tracing decision unit is used to integrate the basic physicochemical parameters collected by the soil multi-parameter sensing array with the quantitative bioactivity index generated by the soil bioactivity index calculation engine, perform multi-level data fusion and model inference, and output a comprehensive quality report including soil health level, limiting factor diagnosis and ecological restoration potential assessment.

2. The soil quality monitoring system for uncultivated wasteland according to claim 1, characterized in that, The soil multi-parameter sensing array consists of multiple heterogeneous sensing probes integrated in a star-shaped topology on the end sensing head of a rigid probe. Each of the sensing probes is evenly distributed in a circle in space, and their sensing end faces are on the same horizontal plane. The heterogeneous sensing probe includes a soil volumetric water content sensor, a soil temperature sensor, a soil pH sensor, a soil electrical conductivity sensor, and a soil redox potential sensor. All the sensor probes are connected to the data acquisition and preprocessing circuit board inside the probe rod via internal shielded cables. The data acquisition and preprocessing circuit board is used to synchronously sample, convert analog signals to digital signals, perform temperature compensation and preliminary filtering on the analog signals output by each sensor, generate a data packet of physicochemical basic parameters with timestamp synchronization, and upload it through the first wireless communication module integrated inside the probe rod.

3. A soil quality monitoring system for uncultivated wasteland according to claim 2, characterized in that, The microbial metabolic response excitation module is integrated at the center of the sensing head of the rigid probe, and includes a micro multi-chamber liquid storage tank, a set of micro precision injection pumps, and a ring microdialysis array. The micro multi-chamber liquid storage tank is independently encapsulated with at least four different types of standardized metabolic substrate concentrates; The number of the micro-precision injection pumps is the same as the number of reservoirs. The inlet of each micro-precision injection pump is connected to the corresponding reservoir through a medical-grade silicone tube, and the outlet is connected to the corresponding independent microchannel on the annular microdialysis array. The annular microdialysis array is made of a biocompatible material with nanoscale pores uniformly distributed on its surface. The annular microdialysis array is arranged coaxially around the central region of the heterogeneous sensing probe. The operation of the microbial metabolic response activation module is triggered by instructions from the central controller. According to the preset activation protocol, the designated micro precision injection pump is driven to inject a standardized metabolic substrate concentrate of a specific type and volume into the corresponding annular microdialysis array microchannel at a constant flow rate.

4. A soil quality monitoring system for uncultivated wasteland according to claim 3, characterized in that, The metabolic response signal capture and analysis module includes a gas metabolite monitoring submodule, a solution ion monitoring submodule, and a thermodynamic monitoring submodule; The gas metabolite monitoring submodule consists of a high-sensitivity carbon dioxide sensor and a methane sensor, which are connected to a miniature gas collection hood via a gas pipeline. The miniature gas collection hood is retractably installed above the probe sensing head and descends and seals the soil surface containing the annular micro-percolation array during the excitation phase, forming a temporary micro-gas chamber. The solution ion monitoring submodule consists of a set of ion-selective field-effect transistor sensors, which are embedded in a specific section of the annular microdialysis array. The thermodynamic monitoring submodule is composed of a high-precision miniature thermopile sensor, whose thermosensitive nodes are in direct contact with the soil micro-domain. The metabolic response signal capture and analysis module is activated at the same time as the microbial metabolic response excitation module is started. It synchronously records gas concentration, ion concentration and temperature change data at a sampling frequency of not less than once per second, and integrates them into a time-aligned multi-dimensional metabolic response kinetic curve data packet, which is then uploaded through the second wireless communication module.

5. A soil quality monitoring system for uncultivated wasteland according to claim 1, characterized in that, The calculation process of the soil biological activity index calculation engine follows these steps: First, a moving average filter is applied to each of the multidimensional metabolic response kinetic curves to remove high-frequency noise, and then the time axis is aligned based on the excitation initiation time. Next, a set of preset time-domain and frequency-domain characteristic parameters are extracted from each preprocessed dynamic curve. The time-domain characteristic parameters include the maximum response intensity of the curve, the time required to reach 50% of the maximum intensity, the total area under the curve, and the time required for the response to decay to 10% of the peak value. The frequency-domain characteristic parameters include the amplitude of the main frequency component obtained by fast Fourier transform. The feature parameters from different monitoring submodules are then input into a pre-trained multilayer perceptron neural network model. The number of input layer nodes in the multilayer perceptron neural network model is equal to the total number of feature parameters. The hidden layer contains two fully connected layers, and the number of output layer nodes is three, corresponding to three core bioactivity indices: basal metabolic activity index, substrate utilization diversity index, and metabolic function stability index.

6. A soil quality monitoring system for uncultivated wasteland according to claim 5, characterized in that, The basal metabolic activity index is determined by the weighted sum of the total area under the carbon dioxide production curve and the maximum response intensity. The substrate utilization diversity index is obtained by calculating the cosine similarity matrix between the feature vectors of response curves to different types of standardized metabolic substrates, and further calculating the information entropy of the cosine similarity matrix. The metabolic function stability index is calculated by analyzing the rate of change of the decay time constant and recovery rate of the microbial response curve when different substrate stimuli are applied continuously or alternately.

7. A soil quality monitoring system for uncultivated wasteland according to claim 1, characterized in that, The soil quality comprehensive evaluation and source tracing decision unit has a built-in soil quality knowledge graph, which stores the complex association rules between soil physicochemical parameters, biological activity index and soil health status, ecological function and limiting factors in the form of triples. The data fusion and model inference process performed by the comprehensive soil quality evaluation and source tracing decision-making unit is as follows: First, the received physicochemical basic parameter data package and the quantitative biological activity index data package are spatiotemporally matched and normalized to form a unified soil feature vector for the current monitoring point. Next, the soil feature vector is input into the inference model based on the coupling of knowledge graph and Bayesian network. The soil feature vector is used as observation evidence to activate the relevant association rules in the soil quality knowledge graph, calculate the posterior probability of different soil health level hypotheses, and determine the level with the highest posterior probability as the soil health level of this monitoring point. At the same time, based on the threshold and logical relationship defined in the soil quality knowledge graph, the main limiting factors that cause the soil health level to not reach the optimal level are diagnosed. Finally, by combining the soil health level, the diagnostic results of the limiting factors, and the time series trend of historical monitoring data, a comprehensive score of the ecological restoration potential of the uncultivated wasteland is calculated using a linear weighted decision model, and a structured comprehensive quality report is generated.

8. A soil quality monitoring system for uncultivated wasteland according to claim 3, characterized in that, The standardized metabolic substrate concentrate includes at least a glucose solution, a protein hydrolysate solution, a cellulose derivative solution, and a phenolic compound solution.

9. A soil quality monitoring system for uncultivated wasteland according to claim 1, characterized in that, The system adopts a hybrid power supply mode that combines solar power supply units with high-capacity lithium-ion battery packs; The solar power supply unit includes a foldable photovoltaic panel and a maximum power point tracking charge controller, which are used to power the system during the day and charge the lithium-ion battery pack. The lithium-ion battery pack is used to provide continuous power at night or in low light conditions.

10. A soil quality monitoring system for uncultivated wasteland according to claim 1, characterized in that, The system is deployed as a networked monitoring system; A single monitoring point constitutes a monitoring node, and multiple monitoring nodes form a wireless sensor network through a self-organizing network protocol; Each monitoring node will aggregate the processed data to the regional gateway node, which will then upload the data to the cloud server via long-distance wireless communication technology.

Images